setup-mofa-remote-tool
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches installation requirements and a specialized Python wheel from 'connect.aiscientist.tools' and the 'mims-harvard/ToolUniverse' GitHub repository. The binary download is explicitly pinned with a SHA-256 hash to ensure integrity.
- [COMMAND_EXECUTION]: Instructs users to perform complex shell operations including environment creation using 'uv' and 'venv', package installation via 'pip', and service management using the 'tu' (ToolUniverse) CLI. These are standard for the tool's deployment.
- [INDIRECT_PROMPT_INJECTION]: The skill defines a 'run_mofa_factors' tool that accepts multi-omics data ('views') as JSON input. While this creates a processing surface for external data, the risk is inherent to the tool's function and the guide provides warnings against sending private data to shared deployments.
- Ingestion points: Data provided to the 'run_mofa_factors' operation via JSON arguments.
- Boundary markers: None explicitly defined in the prompt templates, though JSON structure acts as a basic delimiter.
- Capability inventory: The skill environment allows file writes, package installation, and network communication via the 'tu serve' command.
- Sanitization: Not explicitly documented in the setup guide; relies on underlying tool implementation.
Audit Metadata